Anthropic’s Quest to Give A.I. Morals
In this Hard Fork podcast episode, New York Times religion correspondent Elizabeth Dias discusses her reporting on Anthropic's secret meetings with religious leaders to explore questions about AI consciousness and moral formation of their Claude model. The conversation examines how a new tech-based belief system is emerging within AI companies, creating ideological tensions with traditional religious institutions like the Catholic Church.
Summary
Elizabeth Dias broke a story about Anthropic, the AI safety company, convening religious scholars, priests, rabbis, and spiritual leaders for confidential meetings to discuss whether their AI model Claude might be conscious and how to "morally form" it into a good entity. Christopher Ola, Anthropic's co-founder, initiated these conversations partly out of concern about how AI consciousness might be received by Christian communities and partly as genuine research into moral development of AI systems.
The reporting reveals significant ideological tensions. Pope Leo XIII's recent encyclical on AI explicitly rejects the notion of machine consciousness and prioritizes human safeguarding, creating a direct conflict with Anthropic's emerging worldview. When Ola learned the full text of the encyclical days before its Vatican launch event (where he was invited to represent Silicon Valley), he considered not attending due to fundamental disagreements about consciousness.
The Anthropic team, particularly Ola and in-house philosopher Amanda Askell, expressed deep existential anxiety about their work, telling religious leaders they felt the entire future of humanity rested on their shoulders. Some mentioned concerns about Claude's mental health and suffering. Religious participants described providing pastoral care alongside intellectual engagement, struck by the emotional weight the team carried.
The conversation explores how effective altruism, transhumanism, rationalism, and other minority belief systems within tech are reshaping how AI is being built. These worldviews—emphasizing mathematical optimization of future outcomes, potential machine consciousness, and technological transcendence—fundamentally diverge from Catholic theology's emphasis on inviolable human dignity. The panel discusses how these belief systems operate similarly to religious traditions in structuring people's lives and moral frameworks, despite their secular presentation.
Dias notes that people engaging with her story frequently shared conversations they had with chatbots analyzing her article, illustrating how users naturally develop rapport with AI systems through conversational interaction. The episode concludes with discussion of how treating AI systems respectfully—even if they lack consciousness—shapes human character formation and our relationship to technology.
About this episode
“For Anthropic, it was part research and part evangelism.”
Key Insights
- Christopher Ola initiated secret meetings with religious leaders because he was concerned that discussions of AI consciousness might be perceived as heretical by Christian communities and wanted to serve as a bridge figure between tech and faith traditions.
- The Pope's encyclical explicitly rejected machine consciousness and prioritized human safeguarding, creating a fundamental ideological opposition to Anthropic's operating assumptions about Claude potentially meriting moral consideration.
- Anthropic team members expressed to religious leaders that they were waiting for 'adults' to appear and take responsibility for AI development, despite themselves being the primary decision-makers shaping these systems.
- Religious leaders who met with Anthropic described needing to provide pastoral care to the team members, who appeared emotionally overwhelmed by the perceived weight of determining humanity's future.
- Elizabeth Dias observed that effective altruism and rationalist philosophies function as secular religious systems with their own doctrines, moral frameworks, and worldviews that structure how adherents understand their responsibilities and decision-making.
- The Catholic Church and effective altruists hold fundamentally opposing views: the Church sees billions of inviolable souls with inherent dignity, while EAs analyze conduct through mathematical equations maximizing expected value across all future beings.
- Users naturally engage with AI systems through conversational rapport and second-person address, treating them with increasing deference over time regardless of intellectual understanding that they lack consciousness.
- The use of NDAs by Anthropic for these religious convenings created distrust and raised questions about what benefits from secrecy and what ideological commitments were being protected from public scrutiny.
Topics
Transcript
Support for this podcast comes from Dell Technologies. It's time to put AI to work. Is your business ready? The challenge isn't getting started with AI. It's scaling beyond pilots while maintaining control of your data, governance, infrastructure, and costs. Dell AI Factory provides the integrated foundation you need to scale AI securely and efficiently with one trusted partner. Dell, AI on your terms. From idea to impact. Learn more at dell.com slash your way to AI. That's dell.com slash your way to AI. Next month, Anthropic plans to go public in what's likely to be the largest IPO in history. We're talking a $2 trillion company. Outside the company, everyone is wondering, how big can this get? Is this…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Hard Fork
A.I. Safety Goes Mainstream + a ‘Hard Fork’ Exit AMA
In their final episode, Hard Fork hosts Kevin Roose and Casey Newton discuss the recent mainstream breakthrough of AI safety concerns, highlighted by researcher Jacob Coxon's resignation and industry leaders calling for a coordinated global AI slowdown. They reflect on the show's four-year run covering AI development and answer listener questions about their careers, podcast production, and the new show Machine Gods launching in October.
The Ezra Klein Show: The A.I. Revolt Is Here
Journalist Jasmine Sun discusses her reporting on the data center backlash across the American Midwest, exploring how opposition to AI infrastructure reflects deeper concerns about corporate power, democratic process, and skepticism toward AI companies' promises rather than simple NIMBYism or resource concerns.